Condition & population heart attack relative risks

[Guest guest]

The below papers are pdf-availed.

1. Baccarelli, Emelia J Triggers of MI for the individual and in the community Lancet 2011; 377: 694-6

Myocardial infarction is characterised by an extendedsubclinical phase leading to the sudden onset of life-threatening acute events (figure). Acute myocardialinfarction is often preceded by specific triggers, whichinclude common activities such as alcohol consumption,heavy meals, physical exertion, and stressful events.Most research on myocardial infarction triggers hasemphasised the increased risks to the individual. Thepopulation burden of myocardial infarction triggers hasnot received as much attention.

In The Lancet, Tim Nawrot and colleagues1report a meta-regression analysis of 36 studies on13 myocardial infarction triggers to address this gap.Instead of just focusing on the risk to the individual,these researchers used odds ratios and frequenciesof each trigger to compute population-attributablefractions (PAFs). PAFs estimate the proportion of casesthat could be avoided if a risk factor were removed.2PAFs depend not only on the risk factor strengthat the individual level but also on its frequency inthe community. A very strong but uncommon riskfactor will be very detrimental for the few individualswho are exposed to it, but nonetheless account for arelatively small number of cases in the community.

Nawrot and colleagues skillfully show how theconcepts of individual versus population risks applyto myo cardial infarction triggers. At the individuallevel, cocaine use was by far the strongest trigger(odds ratio 23·7). However, because of the lowreported frequency in the community,3 only 0·9%of myo cardial infarction cases were estimated to betriggered by cocaine.

Conversely, a novel insight provided was the populationburden of air pollution. Particulate matterof 10 μm or less in aerodynamic diameter (PM10)is the air pollutant most consistently associatedwith myocardial infarction onset.4 If the levels ofPM10 in a hypothetical city were to be decreased by30 μg/m³, 4·8% of myocardial infarctions might beavoided or delayed. Despite the low relative risk (oddsratio 1·05) associated with such a change in PM10, thepopulation benefit would be considerable becausethe entire community of this hypothetical city wouldbe advantaged by the reduction in air pollution. The30 μg/m³ PM10 decrease corresponds to a changethat would bridge the gap in several European citiestoward the 20 μg/m³ annual mean limit recommendedby WHO.5 However, this decrease is larger than thatneeded or even achievable in most cities in the USAand several in Europe. For

instance, in 2000,6 mostUS cities had PM10 annual levels below 30 μg/m³ andmany had levels below 20 μg/m³. The PAF of 1·57%computed for a 10 μg/m³ PM10 decrease is still far fromnegligible, and might be a more appropriate estimateof myocardial infarction triggering burden in thosecommunities exposed to low-to-moderate levels of airpollution. Yet the two mitigation scenarios examinedby Nawrot and colleagues are too moderate for mostAsian metropolitan areas, where PM10 levels areconsiderably higher than in the USA and Europe.7

Current research into myocardial infarction triggershas several limitations highlighted by Nawrot andcolleagues. Of the 13 triggers identified, only six wereassessed in multiple studies. The single-study resultsinclude the triggers with highest strength (cocaine,heavy meals), and most of those with the largest PAFs(traffic, alcohol, coff ee). Hence, the studies will need tobe replicated as these high risk–burden estimates couldjust represent the “winner’s curseâ€.8 In addition, severalmyocardial infarction triggers have been loosely defined.Traffic exposure, for instance, might be considereda composite trigger resulting from air pollution,driving-related stress, and noise.

One major area of research that has been understudiedis how triggers contribute to health-care disparities byinteracting with each other and with cardiovascularrisk factors (figure). For instance, the models assume aprevalence of 100% for air pollution, but it is establishedthat air pollution is inordinately concentrated in lowersocioeconomic neighbourhoods.9 Similarly, triggersmight be most relevant to individuals with higher levelsof cardiovascular risk factors, which also are clustereddisproportionately in racial minorities and individualswith lower socioeconomic status.10

Despite limitations, Nawrot and collaborators haveprovided us with an exemplary piece of epidemiologicalwork that furthers our understanding of myocardialinfarction triggers. Their work stands as a warningagainst overlooking the public health relevance of riskfactors with moderate or weak strength that have highfrequency in the community.

1 Nawrot TS, L, Künzli N, Munters E, Nemery B. Public health importance of triggers of myocardial infarction: a comparative risk assessment. Lancet 2011; 732–40.

Abstract

Background

Acute myocardial infarction is triggered by various factors, such as physical exertion, stressful events, heavy meals, or increases in air pollution. However, the importance and relevance of each trigger are uncertain. We compared triggers of myocardial infarction at an individual and population level.

Methods

We searched PubMed and the Web of Science citation databases to identify studies of triggers of non-fatal myocardial infarction to calculate population attributable fractions (PAF). When feasible, we did a meta-regression analysis for studies of the same trigger.

Findings

Of the epidemiologic studies reviewed, 36 provided sufficient details to be considered. In the studied populations, the exposure prevalence for triggers in the relevant control time window ranged from 0·04% for cocaine use to 100% for air pollution. The reported odds ratios (OR) ranged from 1·05 to 23·7. Ranking triggers from the highest to the lowest OR resulted in the following order: use of cocaine, heavy meal, smoking of marijuana, negative emotions, physical exertion, positive emotions, anger, sexual activity, traffic exposure, respiratory infections, coffee consumption, air pollution (based on a difference of 30 μg/m3 in particulate matter with a diameter <10 μm [PM10]). Taking into account the OR and the prevalences of exposure, the highest PAF was estimated for traffic exposure (7·4%), followed by physical exertion (6·2%), alcohol (5·0%), coffee (5·0%), a difference of 30 μg/m3 in PM10 (4·8%), negative emotions (3·9%), anger

(3·1%), heavy meal (2·7%), positive emotions (2·4%), sexual activity (2·2%), cocaine use (0·9%), marijuana smoking (0·8%) and respiratory infections (0·6%).

Interpretation

In view of both the magnitude of the risk and the prevalence in the population, air pollution is an important trigger of myocardial infarction, it is of similar magnitude (PAF 5–7%) as other well accepted triggers such as physical exertion, alcohol, and coffee. Our work shows that ever-present small risks might have considerable public health relevance.

Table 3: Characteristics of the studies on particulate air pollution and non-fatal myocardial infarction---------------------------------------------------------Prevalence of exposure* OR¶ (95% CI) PAF (95% CI)---------------------------------------------------------Air pollution, 10 μg/m³ reduction (n=11)¶ 100% 1·02 (1·01–1·02) 1·57% (0·89 to 2·15)Air pollution, 30 μg/m³ reduction (n=11)¶ 100% 1·05 (1·03–1·07) 4·76% (2·63 to 6·28)Alcohol 3·2% 3·1 (1·4–6·9) 5·03% (2·91 to 7·06)Anger (n=4)¶ 1·5% 3·11 (1·8–5·4) 3·07% (1·19 to 6·16)Cocaine use 0·04% 23·7 (8·1–66·3) 0·90% (0·28 to 2·55)Coff ee 10·6% 1·5 (1·2–1·9) 5·03% (2·08 to 8·71)Emotions positive 1·0% 3·5 (0·7–16·8) 2·44% (–0·30 to 13·64)Emotions negative (n=3)¶ 1·2% 4·46 (1·85–10·77) 3·92% (0·99 to 10·34)Heavy meal 0·5% 7·00 (0·8–66) 2·69% (–0·09 to

23·00)Marijuana 0·2% 4·8 (2·9–9·5) 0·75% (0·38 to 1·67)Physical exertion (n=6)¶ 2·4% 4·25 (3·17–5·68) 6·16% (4·20 to 8·64)Respiratory infection (n=4)¶ 0·4% 2·73 (1·51—4 95) 0·57% (0·17 to 1·29)Sexual activity (n=2)¶ 1·1% 3·11 (1·79–5·43) 2·21% (0·84 to 4·53)Traffic exposure 4·1% 2·92 (2·22–3·83) 7·36% (4·81 to 10·49)--------------------------------------------------------- OR=odds ratio. PAF=population attributable fraction. * Prevalence was based on control time window. It was estimated from the control group (for case-control studies) or the control period (for case-crossover studies). When several studies existed for a same trigger, the average prevalence of the risk factor was calculated by weighting by the sample size of each study. For triggers studied in more than one study, the prevalence was based on the weighted average. ¶ OR based on

pooled OR and prevalence based on weighted means.

-- Al Pater, alpater@... -- Aalt Pater